Saturday, October 25, 2014

Empirical IP Research Conference: copyright

Effect of Copyright Infringement on Creative Incentives

Facilitator: Chris Sprigman (NYU)

Joel Waldfogel (Minnesota Carlson School of Management)
Topic: The strengths and limits of the natural experiment methodology to explore the effects of piracy on both industry output and creative incentives, Commenting on: Brett Danaher & Michael Smith, Gone in 60 Seconds: The Impact of the Megaupload Shutdown on Movie Sales (2013), Christian Peukert et al., Piracy and Movie Revenues: Evidence from Megaupload, A Tale of the Long Tail? (2013)

Do the rewards/incentives created by IP promote welfare? Welfare is consumer surplus plus revenue minus costs of creation/production/distribution.  Incentives depend on profit, not welfare.  Rather than “is stealing sapping our revenue,” ask whether enforcement actions like HADOPI or the Megaupload shutdown stimulate revenue.  Important question for big actors like MPAA, but relevance for creative incentives is indirect.

Both papers use Megaupload shutdown as discontinuous variation on feasibility of copying and ask about effects on legal purchases: not about profits/creation but revenue.  Want a measure of change in movie copying to isolate its impact on purchase. Not observed, so use country as proxy. High-copying countries like Spain provide a large experiment and lower-copying countries like the US provide an implicit control.

But: If revenues and costs are both falling, it’s not clear that digitization sapped incentives. What’s happening in the motion picture industry?  Digital camera costs falling, even for big budget movies.  Huge growth in production.  Supply responses take time; we need to distinguish between “is the Man getting paid?” and “is copyright fulfilling its function?”

Christopher Buccafusco (Chicago-Kent)
Topic: Translating Piracy’s Effect on Sales into Piracy’s Effect on Creative Incentives, Commenting on Commenting on: Brett Danaher & Michael Smith, Gone in 60 Seconds: The Impact of the Megaupload Shutdown on Movie Sales (2013), Christian Peukert et al., Piracy and Movie Revenues: Evidence from Megaupload, A Tale of the Long Tail? (2013)

Deep heterogeneity in the markets; effects on incentives may themselves be variable.  In the second paper, the Megaupload shutdown is pretty good for blockbusters and not helpful/possibly even harmful for smaller movies that rely on word of mouth for success. These aren’t simple or closed systems.  Need to be able to study the effects as they move around. 

Also, incentives for whom?  Who are the producers we care about? That can affect kinds of works created, who creates them, amount that gets invested in various groups, etc. Extrinsic v. extrinsic motivation: complicated relationship between incentive to create and output of products.  These won’t show up in large scale econometric studies and require different methodologies to answer.

What else don’t we know? These aren’t papers about the costs of production, so we need that. There’s not much about overall welfare.  What questions can experimentalists help answer?  Ultimately these are individuals, who might be better studied in labs or qualitative interviews.  To what extent do behavioral biases affect the kinds of creativity we get?

Peter DiCola (Northwestern)
Topic: Generalizability of Studies in One Creative Industry, Commenting on: Rahul Talang & Joel Waldfogel, Piracy and New Product Creation: A Bollywood Story (2014), Joel Waldfogel, Copyright Protection, Technological Change, and the Quality of New Products: Evidence from Recorded Music Since Napster, 55 J.L. & Econ. 715 (2012)

Waldfogel: estimates of quality of new music since Napster, using within rating system variations.  Data from 2000-2007 (end of decade ratings tend to be harsher on most recent music).  Music did not get worse after Napster by various quality measures, including airplay. 

Bollywood: revenue crash and decline in movie production 1985-2000, adjusted for quality.  But DiCola sees a v shape: revenue craters and comes back in 1992-93, making the institutional story more complex.  Unauthorized use: the causal claim comes from a historical narrative, but there’s VHS and cable. VHS proliferation starts in 1985; cable piracy starts in 1992.  Pro-studio change in 1994-95, and enforcement complaints continue through 1997.  Cable piracy starts when things start to recover.  Need a more precise timeline story, in order to generalize from one industry to another and to create a more formal model.

Pamela Samuelson (UC Berkeley)
Topic: Relevance of Empirical Research to Policymakers, Commenting on: National Academies Of Science Report, Copyright in the Digital Era: Building Evidence for Policy (2013)

Need more studies like DiCola’s recent studies of musicians.  Congress/the Copyright Office are willing to hear about real creator communities.  Questions about how to get more people to register copyrights/record transfers are empirical questions.  Are the fees too high? Would differential pricing help?

Katherine Strandburg: Transformativeness as a potential place for empirical data. Courts might be more receptive than Congress.

Samuelson: could identify target audiences for particular work; new audience makes fair use more plausible. Not everything works that way, but some targeted data could help.

McKenna: we never would have had iTunes without Napster.  Second- and third-order effects of legal rules. When you frame “the effects of getting rid of Megaupload” you need to think about the other effects of the legal rule that took down Megaupload, and that can’t be limited to Megaupload.

Talang: industries can recalibrate themselves.  Do we become too quick to offer a regulatory response?  Spotify: new revenue opportunities; labels are appropriating more of that v. the artists.  Will the market be able to work itself out?  India: ultimately people chose quality over free.

Sprigman: porn industry. Used to depend on the feature; now there’s a shift to the live cam/performance, which can’t be copied. 

Talang: growth in concert revenue in the music industry is another example: it’s an experience to be sold. 

Sprigman: industrial organization issue.  Slate article on Mindgeek and pornography—they own content producers and tube sites with a ton of piracy, including of content owned by the company that owns the tube sites.  Piracy takes from one hand and gives to the other.  Piracy drove down the price of the industry players they acquired; innovative in revolting ways.

Lunney: we haven’t mentioned software: is copyright working well there?  If so, why or why not?

DiCola: the calculations about decay in appeal of music over time would also be interesting to do for videogames.  There are people who use old simulators but most are interested in the next edition.

Samuelson: Breyer’s Uneasy Case for Copyright focused on books, but a little on computer programs. Software isn’t one industry but many; different mechanisms for capturing value, a lot not on copyright—complements, first mover advantage, updates, customization, ads.

Sprigman: platforms v. hosted games—different vulnerabilities to piracy. There must be something that the audiences want that sustains the platforms.

Bechtold: DRM on platforms makes it inconvenient to play pirated Xbox games; inconvenience and time constraints make buying easier, and interactive/online elements are also harder w/pirated or even secondary market purchases.

Strandburg: videogames are interesting because of how people consume them/potential substitutability with other works of entertainment that one might consume, e.g. board games.  What happens when you don’t have certain kinds of protections?

Copyright breakout session: Investigating Copyright Infringement Standards
Facilitators: Ben Depoorter (UC Hastings) & Paul Heald (Illinois)

Depoorter: Cognitive biases may affect infringement inquiries: confirmation bias, jury group effects, hindsight bias, etc. Applications in music: gaps between music theory and law. Some papers argue that music is different, similar story told about pictures (that’s me!).  Auditory sensation, time, musical elements lay observers find hard to detect; musical language is intuited but not understood; still we ask lay listeners to make the determination, but they consider performance characteristics rather than compositional ones.  Jamie Lund’s interesting empirical work: lots of room for false positives given how the cases are litigated.  What is to be done?  ABKCO v. Harrisongs: students look for similarities in My Sweet Lord and He’s So Fine; judge did it right by identifying similarities in musical forms.  Improvements: Lemley suggests increasing the role of experts in improper appropriation test; Pam Samuelson suggested reversing the test and looking first for improper appropriation and then to copying.

Heald: Laroche: mathematical approach to distance between a piece of music and an allegedly infringing piece—how many moves do you have to make to get from one to another.  When they map distance results to songs on the Columbia/USC music infringement database, they don’t get much consistency with the infringement findings, but that doesn’t mean their model is wrong.

For further research: redo Lund’s experiment with composers instead of lay listeners.  If © is a tort, we need a model for determining whether someone w/ a property right has suffered actionable damage.  What we ought to care about is whether composers are behaving reasonably when they borrow from other composers.  Reasonableness doesn’t require perfect rationality, just whether the appropriator is within the normal bounds of musical appropriation norms. 

We can also test for what composers think the law is.  A lot of compositional norms are distorted by an erroneous perception of the law.  Organists: is it ok for them to buy a piece of music and copy it to lay it out on the keyboard so they don’t have to turn pages since they are busy with hands and feet? Some think they can’t do that so they buy 8 copies.

Samuelson: role of experts.  Long period when experts were considered inappropriate for certain issues; are experts good/bad idea?  Also ideas about getting courts to move away from total concept and feel.

Jeanne Fromer: norms within areas differ—appropriation artists think differently than other groups. Experts about what?  Norms among artists; consumer understanding; something else?  Collecting data is hard because it’s hard to tell whether courts or judges are doing the same thing, or even the same thing from case to case—it’s not even a multifactor test.

Beebe: cases are borderline uncodable. Confirms slipperiness of concept of similarity in human cognition. 

Heald: efficient results require incredible flexibility of the infringement test in order to deal with the otherwise huge/rigid rights granted by a valid copyright; thus he’s not confident that much rationalization is possible.

Note that computer software doctrine is much more deferential to how programmers behave; not the lay observer. Would like to see that notion in all copyright law. Why care what lay auditor/observer sees instead of how a reasonable actor behaves towards relevant others?

Irina Manta: market substitution would still be relevant.  What about a study of copyright infringement surveys in TM infringement style?  What do people think about similarity in works and does it correlate with results in litigated cases?  If there’s a disparity we’d then want to know why/which we thought were right.

DiCola: The harm comes from the consumers—it’s either the primary part of the case or the result.  Heald is saying that the breach comes from D’s behavior, but harm has to be connected in some way to the wrongful behavior.

Samuelson: may need to distinguish reproductions from derivative works. Infringement test right now gets used the same between reproduction/derivative work.  Students ask what the difference is and she tries to give them a set of factors, but it’s difficult. Castle Rock: implausible to view SAT case as a case about substantial similarity and reproduction. TV series is not a quiz.  If it’s plausible at all, it’s a derivative works right case, and she doesn’t find it very convincing even as such. Good teaching case for discussing the difference.

DePoorter: if jury is substitute for survey, it’s harder to debias juries. Could try to mimic lab setting; valuable contributions would be in manipulating debiasing interventions.

Manta: currently parties are running their own unscientific surveys on the jurors. It would be fairer to do more; it doesn’t take much to skew people’s perceptions of similarity in our studies.  In a lab study you wouldn’t know anything about the litigation, who the parties are, etc.

DePoorter: what about not telling subjects that the D copied?

Manta: telling them that there was copying had a clear effect, as did telling them that a lot of labor went into the original; telling them there was market substitution did not have an effect. Subjects care about labor even though they’re not supposed to.

McKenna: even if people say works are similar, you have to figure out what’s triggering the similarity assessment, and a lot of things are supposed to be irrelevant. Controls are used in TM, at least theoretically. © control is harder to figure out.

Manta: true; our studies looked at smaller works like images rather than books/movies.

McKenna: Gucci v. Guess—sued over diagonal arrangement of overlapping Gs, beige on beige; Gucci claims all this as its trade dress, even though diagonals and beige on beige are common. Control: yellow on blue, arranged in square, with nonstylized letters—they changed all the elements, and the judge still said that was a fine control. Hard to have confidence in good © controls.

DiCola: evidence from criminal law suggests that debiasing is incredibly difficult/practically impossible.  Getting people not to think about labor may be impossible—how do we change their minds? 

Manta: but surveys help with that because people surveyed aren’t being told about labor; when it gets to the courtroom, you should limit what you allow attorneys to say.  Then it would be harder to ignore the survey evidence.

Fromer: courts have ignored this by sending it to jury.  Forcing the court’s hand by bringing in evidence and requiring its confrontation—in patent, inventors are forced to articulate what they think they’ve created up front. In copyright, we don’t have that, and there are good reasons, but it means that when you get to litigation you haven’t articulated what’s important about this work up front. We need a point in the © system as in patent and TM for forcing articulation of features P considers important.

No comments: